Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Fix Linear layer bias gradient computation; add size checks to CUDA functions #170

Merged
merged 4 commits into from
Aug 15, 2022
Merged

Fix Linear layer bias gradient computation; add size checks to CUDA functions #170

merged 4 commits into from
Aug 15, 2022

Conversation

hweom
Copy link
Contributor

@hweom hweom commented Jul 28, 2022

What does this PR accomplish?

  • 🩹 Bug Fix

Closes #169

Changes proposed by this PR:

  1. Add an assert to CUDA copy() function to check that source and destination have the same size.
  2. Add an assert to CUDA gemm() function to check that computed matrix multiplication dimensions match the passed tensor sizes.
  3. Fix the Linear layer bias gradient computation by summing all gradients in the batch instead of copying the output gradient to the bias gradient (which is incorrect since the output gradient contains N items, not 1).

After this change, cuda-memcheck no longer finds errors.

Notes to reviewer:

All unit tests continue to pass except ui which is also broken at HEAD.

📜 Checklist

  • The juice-examples run just fine

drahnr
drahnr approved these changes Jul 28, 2022
View reviewed changes
Copy link
Member

@drahnr drahnr left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good, the asserts are a good addition, we might want to relax them just a little bit to allow for more niche use cases.

coaster-blas/src/frameworks/cuda/helper.rs Show resolved Hide resolved
juice/src/layers/common/linear.rs Outdated Show resolved Hide resolved
drahnr
drahnr approved these changes Aug 4, 2022
View reviewed changes
Copy link
Member

@drahnr drahnr left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The only thing that fails is the ui tests. Other than that, LGTM!

@hweom
Copy link
Contributor Author

hweom commented Aug 11, 2022

Test fixes are in #172.

@drahnr
Copy link
Member

drahnr commented Aug 11, 2022

Could you rebase this PR onto master? I'll merge right away then

@hweom
Copy link
Contributor Author

hweom commented Aug 14, 2022

Done.

@drahnr drahnr merged commit 6952a49 into fff-rs:master Aug 15, 2022
drahnr added a commit that referenced this pull request Sep 14, 2022
@hweom @drahnr
Implement Sequential layer for the new architecture (#168)
8df57ab 
* Implement Sequential layer

* Fix coaster UI tests (rustc error messages changed in 1.62 (#172)

Co-authored-by: Mikhail Balakhno <{ID}+{username}@users.noreply.github.com>

* Fix Linear layer bias gradient computation; add size checks to CUDA functions (#170)

* Assert the correct tensor sizes in copy() and gemm(); fix related Linear logic

* Check output matrix dims in GEMM; fix corresponding Linear layer logic

* Update coaster-blas/src/frameworks/cuda/helper.rs

Co-authored-by: Bernhard Schuster <[email protected]>

Co-authored-by: Mikhail Balakhno <{ID}+{username}@users.noreply.github.com>
Co-authored-by: Bernhard Schuster <[email protected]>

* More ergonomic net creation and fallible Sequential constructor

* Fix merge mistake in commit 6952a49

Co-authored-by: Mikhail Balakhno <{ID}+{username}@users.noreply.github.com>
Co-authored-by: Bernhard Schuster <[email protected]>
hweom added a commit to hweom/juice that referenced this pull request Feb 25, 2024
@hweom @drahnr
Implement Sequential layer for the new architecture (fff-rs#168)
e658837 
* Implement Sequential layer

* Fix coaster UI tests (rustc error messages changed in 1.62 (fff-rs#172)

Co-authored-by: Mikhail Balakhno <{ID}+{username}@users.noreply.github.com>

* Fix Linear layer bias gradient computation; add size checks to CUDA functions (fff-rs#170)

* Assert the correct tensor sizes in copy() and gemm(); fix related Linear logic

* Check output matrix dims in GEMM; fix corresponding Linear layer logic

* Update coaster-blas/src/frameworks/cuda/helper.rs

Co-authored-by: Bernhard Schuster <[email protected]>

Co-authored-by: Mikhail Balakhno <{ID}+{username}@users.noreply.github.com>
Co-authored-by: Bernhard Schuster <[email protected]>

* More ergonomic net creation and fallible Sequential constructor

* Fix merge mistake in commit 6952a49

Co-authored-by: Mikhail Balakhno <{ID}+{username}@users.noreply.github.com>
Co-authored-by: Bernhard Schuster <[email protected]>
hweom added a commit to hweom/juice that referenced this pull request Mar 10, 2024
@hweom @drahnr
Implement Sequential layer for the new architecture (fff-rs#168)
ec796ac 
* Implement Sequential layer

* Fix coaster UI tests (rustc error messages changed in 1.62 (fff-rs#172)

Co-authored-by: Mikhail Balakhno <{ID}+{username}@users.noreply.github.com>

* Fix Linear layer bias gradient computation; add size checks to CUDA functions (fff-rs#170)

* Assert the correct tensor sizes in copy() and gemm(); fix related Linear logic

* Check output matrix dims in GEMM; fix corresponding Linear layer logic

* Update coaster-blas/src/frameworks/cuda/helper.rs

Co-authored-by: Bernhard Schuster <[email protected]>

Co-authored-by: Mikhail Balakhno <{ID}+{username}@users.noreply.github.com>
Co-authored-by: Bernhard Schuster <[email protected]>

* More ergonomic net creation and fallible Sequential constructor

* Fix merge mistake in commit 6952a49

Co-authored-by: Mikhail Balakhno <{ID}+{username}@users.noreply.github.com>
Co-authored-by: Bernhard Schuster <[email protected]>
drahnr added a commit that referenced this pull request Mar 15, 2024
@hweom @drahnr
@dependabot @opfromthestart
Sync arch refactor (#201)
1eeb939 
* Fix coaster UI tests (rustc error messages changed in 1.62 (#172)

* Fix Linear layer bias gradient computation; add size checks to CUDA functions (#170)

* Assert the correct tensor sizes in copy() and gemm(); fix related Linear logic

* Check output matrix dims in GEMM; fix corresponding Linear layer logic

* Update coaster-blas/src/frameworks/cuda/helper.rs

* Fix merge mistake in commit 6952a49 (#173)

* doc: clarify remote test (#175)

* bump rust-bindgen to 0.60.1, bump cargo lock file (#174)

* build(deps): bump capnp from 0.14.9 to 0.14.11 (#179)

Bumps [capnp](https://github.com/capnproto/capnproto-rust) from 0.14.9 to 0.14.11.
- [Release notes](https://github.com/capnproto/capnproto-rust/releases)
- [Commits](capnproto/capnproto-rust@capnp-v0.14.9...capnp-v0.14.11)

---
updated-dependencies:
- dependency-name: capnp
  dependency-type: direct:production
...

* build(deps): bump tokio from 1.21.0 to 1.23.1 (#183)

Bumps [tokio](https://github.com/tokio-rs/tokio) from 1.21.0 to 1.23.1.
- [Release notes](https://github.com/tokio-rs/tokio/releases)
- [Commits](tokio-rs/tokio@tokio-1.21.0...tokio-1.23.1)

---
updated-dependencies:
- dependency-name: tokio
  dependency-type: direct:production
...

* build(deps): bump bumpalo from 3.11.0 to 3.12.0 (#187)

Bumps [bumpalo](https://github.com/fitzgen/bumpalo) from 3.11.0 to 3.12.0.
- [Release notes](https://github.com/fitzgen/bumpalo/releases)
- [Changelog](https://github.com/fitzgen/bumpalo/blob/main/CHANGELOG.md)
- [Commits](fitzgen/bumpalo@3.11.0...3.12.0)

---
updated-dependencies:
- dependency-name: bumpalo
  dependency-type: indirect
...

* build(deps): bump tokio from 1.23.1 to 1.24.2 (#191)

Bumps [tokio](https://github.com/tokio-rs/tokio) from 1.23.1 to 1.24.2.
- [Release notes](https://github.com/tokio-rs/tokio/releases)
- [Commits](https://github.com/tokio-rs/tokio/commits)

---
updated-dependencies:
- dependency-name: tokio
  dependency-type: direct:production
...

* Now also saves bias layers (#193)

* build(deps): bump openssl from 0.10.41 to 0.10.48

Bumps [openssl](https://github.com/sfackler/rust-openssl) from 0.10.41 to 0.10.48.
- [Release notes](https://github.com/sfackler/rust-openssl/releases)
- [Commits](sfackler/rust-openssl@openssl-v0.10.41...openssl-v0.10.48)

updated-dependencies:
- dependency-name: openssl
  dependency-type: indirect
...

Signed-off-by: dependabot[bot] <[email protected]>

* Do not pass batch_size to cudnnGetRNNParamsSize().

* Add a feature for deterministic (pseudo)randomizing.

* New network architecture pieces: Layer, Descriptor, Context, Network (#165)

* New network architecture pieces: Layer, Descriptor, Context, Network

* Update juice/src/net/descriptor.rs

* Implement Sequential layer for the new architecture (#168)

* Implement Sequential layer

* Fix coaster UI tests (rustc error messages changed in 1.62 (#172)

* Fix Linear layer bias gradient computation; add size checks to CUDA functions (#170)

* Assert the correct tensor sizes in copy() and gemm(); fix related Linear logic

* Check output matrix dims in GEMM; fix corresponding Linear layer logic

* Update coaster-blas/src/frameworks/cuda/helper.rs

* More ergonomic net creation and fallible Sequential constructor

* Fix merge mistake in commit 6952a49

* Add a few more layers to the new architecture (#176)

* Add trainer subsystem with SGD and Adam optimizers (#177)

* Coaster convolution API cleanup (#178)

* Move Convolution workspace into context

* Implement Convolution, Dropout and Pooling layers (#180)

* Move Convolution workspace into context

* Formatting fixes

* Fixed unit tests

* Partial implementation of the Convolution layer

* Implement the remaining parts for Convolution layer

* Implement dropout and pooling layers

* Fix CUDA tensor descriptor size error and adjust layer testing infra

* Extended debug output for layers with custom Debug impl

* Add softmax layers and convert MNIST example (#184)

* Move Convolution workspace into context

* Formatting fixes

* Fixed unit tests

* Partial implementation of the Convolution layer

* Implement the remaining parts for Convolution layer

* Implement dropout and pooling layers

* Fix CUDA tensor descriptor size error and adjust layer testing infra

* Extended debug output for layers with custom Debug impl

* Changed mnist example to the new architecture

* Plumbed the momentum arg in the mnist example

* Implemented softmax and logsoftmax layers

* Remove unnecessary NLL parameter and fix mnist example

* Fix native backend softmax and logsoftmax grad computation

* Changed slicing syntax in native backend softmax functions

* Convert juice benchtests to Criterion (#192)

* Convert Juice benchmarks to Criterion

* Add newline at the end of Cargo.toml

* Made Layer operations return a Result (#186)

* Made Layer operations return a Result

* Change LayerError to contain Boxes

* Update benchmarks for new layer API

* Simplify new_rnn_config()

Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: Mikhail Balakhno <{ID}+{username}@users.noreply.github.com>
Co-authored-by: Bernhard Schuster <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: opfromthestart <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@drahnr drahnr drahnr approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

cuda-memcheck: "Address ... is out of bounds"
2 participants
@hweom @drahnr